基于BP神经网络的多变量海水环境中冷喷涂铜复合防腐防污涂层的铜渗出率预测系统  

Prediction System of Copper Leaching Rate of Cold Spray Cu Composite Antifouling and Anticorrosion Coatings in Multivariate Marine Environment Based on BP Artificial Neural Network

作　　者：丁锐[1] 蒋健明[1] 桂泰江[1] 李相波 

机构地区：[1]海洋涂料国家重点实验室,海洋化工研究院有限公司,山东青岛266071 [2]海洋腐蚀与防护重点实验室,中船重工七二五所,山东青岛266101

出　　处：《中国涂料》2016年第6期46-52,共7页China Coatings

摘　　要：实验应用Matlab 2013b建立了用于冷喷涂Cu-Cu_2O涂层在多环境变量海水中的铜渗出率预测的BP人工神经网络系统。网络采用5-28-1三层网络结构,隐层传递函数为对数S型函数(logsig),输出层传递函数为线性函数(purelin),训练函数采用trainlm函数,训练误差目标设置为1×10^(-7),训练前对网络的权值和阈值进行零初始化。实验结果表明该神经网络具有很好的预测能力和泛化能力,能够有效地预测冷喷涂Cu-Cu_2O涂层在多环境变量海水中的铜渗出率,预测误差在10%以内。BP artificial neural network system used to predict copper leaching rate of cold spray Cu-Cu2O coatings in multivariate marine environment was established by Matlab 2013 b. The network was a three-tier structure of 5-28-1. Hidden layer transfer function was S-shaped logarithmic function（logsig） and the transfer function of the output layer was a linear function（purelin）. The training function token trainlm function and training error target was set to 1×10^-7. Before training, the weights and threshold of the network was initialized to zero. Experimental results indicated that the neural network had good predictive ability and generalization ability. It can effectively predict the copper leaching rate of cold spray Cu-Cu2O coating in multivariate seawater environment and the prediction error is within 10%.

关 键 词：BP神经网络 冷喷涂 Cu-Cu2O涂层 铜离子渗出率 防污 

分 类 号：TQ630.71[化学工程—精细化工]